Twitter and Facebook have been the subject of countless studies and think pieces exploring their use in political uprisings, notably during the Arab Spring. But since that time, Instagram has grown up as a powerful platform in its own right with more than 200 million users across the globe and has been largely ignored.
A new analysis led by computer scientist Lev Manovich, a professor at CUNY's The Graduate Center, changes that focus by diving deep into 13,208 images shared by 6,165 users from the central part of Kiev during several days of the "Maidan" revolution this past February.
News reports tend to feature social media imagery associated with revolution, often by following a hashtag or specific accounts. What’s most interesting about the visualization is that it looks at everyone in and around Maidan (or Independence) Square, not just the protesters. The result is a fascinating juxtaposition of protest imagery—the images of fires, angry crowds, and protest signs—with your typical selfies and snapshots of pretty blue skies.
"You have the everyday and the exceptional. They co-exist, but in moments, the exceptional kind of takes over," says Manovich, who has specialized in visualizing large Instagram data sets from different cities (see past projects we’ve covered, here and here).
The project began in late January, when Manovich, a native of Russia who left as a political refugee, turned a Instagram geographic analysis tool he and his collaborators had been developing to the unrest in Ukraine he’d been following in the news. On Instagram, looking zoomed out at all photos from the area, Manovic says it was "fascinating" and "strange" that, for weeks, it was not obvious that anything political was happening. (In the large visualization, the most notable feature is the six light to dark "waves" that mark day and night).
That changed the evening of February 18, when government forces attacked Maidan Square. Zooming in on the visualization, you can the events clearly. By the next morning, however, again, the unrest was harder to spot. Unusual activity around Maidan Square made up only about one-quarter of images at any time over the approximately five days the revolution took place.
"This gives us such a different representation of the city—what people actually feel," says Manovich, who collaborated with University of California, San Diego, research scientist Mehrdad Yazdani, CUNY art history student Alise Tifentale, and web developer Jay Chow on the work.
Even focusing on Maidan hashtags (in Ukrainian, Russian, and English), while most images are related to the protest movement a few are not—such as the selfie in column 1, row 4 below. In these images, the highest frequency of subjects of images was "crowds," then "fire or smoke," "flags," "other subjects," "portraits," and "barricades."
To Manovich, Instagram is notable because it is still a smaller visual platform. The images are less dominated by powerful accounts, especially government accounts wishing to spread propaganda. In the visualization, which was created by downloading 10 million images a day from Instagram’s public API and "telescoping" to those tagged in the Central Ukraine, each users was responsible, on average, for only one to two photos.
He sees social media as its own reality, one that is distinct from the reality on the ground but nevertheless now fundamentally part of the fabric of revolution. "Depending on how you organize these pictures, you can present a very different picture. So what does it mean to construct a narrative out of this data?" he asks. The sprawling project shows, at the very least, it depends on where and how you focus.
The project website includes many more visualizations and analysis, so you can explore more here. The work will be presented at the second workshop on Big Humanities Data, held in conjunction with the IEEE Big Data Conference later in October.